Fluorescence Enhancement by AuNanostructures: Nanoshells andNanorodsRizia Bardhan,†,§ Nathaniel K. Grady,‡,§ Joseph R. Cole,‡,§ Amit Joshi,� and Naomi J. Halas†,‡,§,*†Department of Chemistry and ‡Department of Electrical and Computer Engineering, §Laboratory for Nanophotonics, Rice University, Houston, Texas 77005, and�Department of Radiology, Baylor College of Medicine, Houston, Texas 77005

Fluorescence imaging has seen wide-spread use in clinical diagnosis andmonitoring processes in biological

systems.1 The development of contrastagents, such as fluorescent probes with en-gineered biomarker functionalities, has be-come integral to the advancement of newbioimaging technologies.2 Fluorescent mol-ecules emitting at wavelengths in the physi-ologically relevant “water window”(700�900 nm) are of particular interestdue to the large penetration depth of near-infrared (NIR) light in most biological me-dia and offer the potential for imaging atsignificant depths in living tissues.3 How-ever, achieving bright fluorescent emissionwith photostable and biocompatiblenear-IR fluorophores has proven to be ex-tremely difficult. It has long been knownthat, in the proximity of a metallic surface,fluorescence emission of molecules can beenhanced; this is also the case for metallicnanostructures and nanoparticles adjacentto a fluorophore.4�7 The presence of anearby metallic nanoparticle can not onlyenhance the quantum yield but also stabi-lize adjacent fluorophores against photo-bleaching, further enhancing their practicaluse in bioimaging applications.8 In new andemerging light-assisted therapeutic appli-

cations such as photothermal cancertherapy, the addition of bright near-IR fluo-rescence to a therapeutic nanostructurecomplex could provide additional diagnos-tic imaging capabilities for this treatmentstrategy that could facilitate clinical use. Un-derstanding precisely how metallic nano-structures enhance molecular fluorescenceis of general fundamental interest and mayultimately provide practical routes to en-hancing light emission from a variety of ma-terials systems and devices far beyond thespecific application of bioimaging.

Metal nanostructures exhibit remark-able optical properties due to excitation oftheir surface plasmons by incident light,which results in a significant enhancementof the electromagnetic field at the nanopar-ticle surface. This enhanced near field canbe used to design highly sensitive chemi-cal and biosensors with specific plasmonresonances tailored by the nanoparticlegeometry.9,10 Metallic nanoparticles havebeen shown to enhance the fluorescenceemission and decrease the molecularexcited-state lifetimes of vicinal fluoro-phores. The fluorescence enhancement isattributable to a combination of processesincluding enhanced absorption by the mol-ecule, modification of the radiative decayrate of the molecule, and enhanced cou-pling efficiency of the fluorescent emissionto the far field.8,11

The plasmon resonant properties of me-tallic nanoparticles can be controlled by op-timizing the nanoparticle topology, dimen-sions, and composition.12�15 Whenplasmonic nanoparticles are much smallerthan a wavelength of light, they are absorb-ers, much like molecules. Nanoparticleslarger than a few tens of nanometers bothabsorb and scatter light. While both the ab-sorption and scattering cross sections of a

*Address correspondence tohalas@rice.edu.

Received for review January 2, 2009and accepted February 11, 2009.

Published online February 20, 2009.10.1021/nn900001q CCC: $40.75

© 2009 American Chemical Society

ABSTRACT Metallic nanoparticles influence the quantum yield and lifetime of adjacent fluorophores in a

manner dependent on the properties of the nanostructure. Here we directly compare the fluorescence

enhancement of the near-infrared fluorophore IR800 by Au nanoshells (NSs) and Au nanorods (NRs), where human

serum albumin (HSA) serves as a spacer layer between the nanoparticle and the fluorophore. Our measurements

reveal that the quantum yield of IR800 is enhanced from �7% as an isolated fluorophore to 86% in a

NSs�HSA�IR800 complex and 74% in a NRs�HSA�IR800 complex. This dramatic increase in fluorescence

shows tremendous potential for contrast enhancement in fluorescence-based bioimaging.

KEYWORDS: fluorescence enhancement · IR800 · gold nanoshells · goldnanorods · frequency domain lifetime decay

ART

ICLE

VOL. 3 ▪ NO. 3 ▪ BARDHAN ET AL. www.acsnano.org744

plasmon resonant nanoparticle increase with increas-ing particle size, scattering begins to dominate in thislarger-size regime.16 Au nanorods (NRs) and Aunanoshells (NSs) are both particularly useful for biologi-cal applications since their near-field and far-field opti-cal properties can be tuned controllably throughout theNIR water window by varying their geometry. For NRs,the aspect ratio defines two distinct plasmon resonancefrequencies associated with the longitudinal and trans-verse dimensions of the nanostructure.17 For NSs, sym-metric spherical nanoparticles consisting of a dielectriccore and a metal shell, the plasmon resonances are de-termined by the relative size of the core and the metalshell layer.18 NSs can be fabricated both in a small sizeregime (d � 50 nm) using Au2S/Au and sizes up to a mi-cron in silica/Au core/shell structures, accessing near-IRresonances over a large size range.16 NRs tuned to thesame NIR plasmon resonance frequency as silica/Au NSsare significantly smaller in size than the correspondingnanoshell, resulting in a much smaller contribution ofscattering to the overall extinction cross section. Study-ing fluorescence enhancement by these two types ofstructures allows us to examine how the properties ofboth nanostructures contribute to this effect.

We have recently shown that the fluorescence ofNIR fluorophores can be significantly enhanced whenthe molecules are in close proximity to NSs.6,7 Fluores-cence enhancements as large as 50 fold were obtainedwhen the plasmon resonance of the nanoshell wastuned to the emission wavelength of the fluorophore,for a nanoshell designed with a significant scatteringcross section at that wavelength. In the study, we com-pare the fluorescence enhancement of the NIR fluoro-phore IR800 positioned a small, controlled distancefrom NSs and NRs. We quantified the fluorescence en-hancement of IR800 experimentally by measuring theangle resolved scattering intensities, frequency domainfluorescence decay, and quantum yield of IR800 be-fore and after binding to the nanoparticles. We also cal-culated the scattering cross sections of NSs and NRs todetermine the scattering intensities of the nanoparti-cles and relative contribution of nanoparticle scatter-ing toward molecular fluorescence enhancement.

When a fluorophore is adsorbed directly onto ametal surface, its fluorescence is quenched. However,at a distance of a few nanometers from the nanoparti-cle surface, the fluorescence can be strongly en-hanced.19 IR800 molecules were placed in close proxim-ity of NSs and NRs surfaces by coating the nanoparti-cles with a 5�11 nm layer of human serum albumin(HSA) in order to prevent quenching of IR800. Serum al-bumin, a large multidomain protein relevant to manyphysiological functions,20 has been conjugated to Aunanoparticles extensively for cell-targetingapplications.21,22 It binds to Au by electrostatic attrac-tion between the amine groups of the protein and thenegative charge on the gold surface or, alternatively, by

covalent attachment between the Au surface andamino acid functional groups present in the protein.23

In this study, HSA acts as both a spacer layer as well asa linker of the fluorophore to the nanoparticle. It con-tains an abundance of �-amines at N-terminals and�-amines on lysine side chains, which covalently bindto IR800 via the N-hydroxysuccinimide (NHS) estergroup to form a stable protein�dye complex.24 TheNHS ester group facilitates protein conjugation, and thepresence of the negatively charged sulfonate groupsmakes IR800 soluble in aqueous media. While theseproperties make IR800 a suitable dye for bioimaging,25

it is limited by low quantum efficiency compared tocommonly used visible dyes such as boron dipyr-romethene (BODIPY) and fluorescein isothiocyanate(FITC). Therefore, by binding IR800 to metal nanoparti-cles with a suitably determined spacer layer, the quan-tum yield can be significantly enhanced.

RESULTS AND DISCUSSIONThe absorption�emission profile of IR800 conju-

gated to HSA in sodium phosphate buffer (Na3PO4, pH� 8.0) is shown in Figure 1a. The absorbance maximumof the protein�fluorophore complex is observed at782 nm, and the fluorescence emission maximum at804 nm. The IR800 molecules were conjugated to HSA

Figure 1. (a) Absorption�emission profile (�max-Ab � 782 nmand �max-Em � 804 nm) of IR800 conjugated with HSA. Chemi-cal structure of IR800 is provided as an inset. (b) Schematicdiagram illustrating the conjugation of HSA�IR800protein�fluorophore complex to nanoshells (NSs) andnanorods (NRs) to form NSs�HSA�IR800 andNRs�HSA�IR800, respectively. The blue curved line repre-sents HSA, and the red dots represent IR800 dye.

ARTIC

LE

www.acsnano.org VOL. 3 ▪ NO. 3 ▪ 744–752 ▪ 2009 745

to form the HSA�IR800 complex, which was thenbound to NSs (NSs�HSA�IR800) and NRs(NRs�HSA�IR800) as demonstrated in Figure 1b.

HSA�IR800 conjugation was performed at roomtemperature, and the protein�fluorophore complexwas dialyzed for 24 h to remove excess dye (see Meth-ods, section II). It was determined that approximately 16molecules of IR800 were bound to each HSA molecule(see Methods, section V). The bioconjugation of theHSA�IR800 complex to the nanoparticles was per-formed in Na3PO4 buffer at room temperature and pro-tected from light, and excess protein�fluorophore mix-ture was removed by centrifuging the nanoparticleconjugates in small aliquots (see Methods, section III).We determined that �63 HSA�IR800 complexes wereconjugated to each NS and �4 HSA�IR800 complexeswere bound to each NR (see Methods, section V). How-ever, the total number of HSA�IR800 complexes bound

to NSs or NRs was kept constant by adjusting the nano-particle concentration, such that the total surface areaavailable for conjugation was equivalent for both NSsand NRs (see Methods, section IV).

The geometry of the nanoparticles coated withHSA�IR800 was characterized with transmission elec-tron microscopy (TEM). TEM micrographs of NSs with di-mensions [r1, r2] � [63, 78] nm, where r1 is the radiusof the silica core and r2 is the radius of the Au shell, areshown in Figure 2a. NRs, [w, l] � [11, 46] nm, where w isthe width and l is the length, are shown in Figure 2c.The NSs and NRs coated with a uniform 8 � 3 nm layerof HSA are shown in Figure 2b and Figure 2d,respectively.

The plasmon resonance of NSs and NRs shifts tolonger wavelengths by 5�8 nm when HSA�IR800 is ad-sorbed, due to the higher refractive index of HSA thanH2O26,27 (Supporting Information Figure S1). The plas-mon resonances of the nanoparticles were tuned to theemission wavelength of the fluorophore�protein com-plex to maximize the fluorescence enhancement.6 Theextinction maximum of NSs�HSA�IR800 was observedat �805 nm, and that of NRs�HSA�IR800 was ob-served at �804 nm, as shown in Figure 3a. The corre-sponding emission spectra of HSA�IR800 conjugatedto the nanoparticles are shown in Figure 3b.HSA�IR800 was used as the control sample ratherthan IR800 in aqueous solution to ensure that the fluo-rophore was in essentially the same chemical environ-ment in all measurements. The emission spectra of NSscoated with HSA without IR800 and NRs coated withHSA without IR800 are also provided in the SupportingInformation Figure S3. The nanoparticles exhibited neg-ligible fluorescence in the near-infrared without the flu-orophore. The reference sample as well as the nanopar-ticle conjugates were excited at 780 nm, and thefluorescence spectra were collected in solution underidentical excitation and detection conditions, allowingdirect comparison of the variousnanoparticle�fluorophore complexes. A maximum

fluorescence enhancement of �40 was measuredfor NSs�HSA�IR800 relative to the control sample,and an enhancement of �9 was found forNRs�HSA�IR800. The emission spectra of thefluorophore�nanoparticle conjugates excited at720 nm are also provided in the Supporting Infor-mation Figure S4. No notable difference in the lineshape of the emission spectra of IR800 in the pres-ence or absence of the nanostructures was ob-served either for excitation at 720 nm or for excita-tion at 780 nm.

Since the nanoparticles used in this experi-ment vary in size, their relative scattering cross sec-tions also differ significantly, which directly affectstheir fluorescence enhancement efficiency. We de-termined the scattering intensities of the NSs andNRs by calculating their scattering cross sections

Figure 2. TEM micrographs of (a) NSs [r1, r2] � [63, 78] nm, (b) NSscoated with 8 � 3 nm HSA�IR800, (c) NRs [w, l] � [11, 46] nm, and(d) NRs coated with 8 � 3 nm HSA�IR800.

Figure 3. (a) Extinction spectra of (i) NSs�HSA�IR800, �max � 805 nm, and (ii)NRs�HSA�IR800, �max � 804 nm are shown. Spectra are offset for clarity. (b)Fluorescence (FL) spectra, �max � 804 nm of (i) NSs�HSA�IR800, (ii)NRs�HSA�IR800, and (iii) control sample HSA�IR800 are shown.

ART

ICLE

VOL. 3 ▪ NO. 3 ▪ BARDHAN ET AL. www.acsnano.org746

using experimentally obtained nanoparticle sizes. Thecalculated scattering intensities for NSs [r1, r2] � [63, 78]nm in aqueous medium were obtained using Mietheory, and the maximum plasmon resonance peakwas observed at 800 nm. The theoretical scattering in-tensity of NRs was calculated by using the finite ele-ment method (FEM) (as implemented by COMSOL Mul-tiphysics). The nanorods were modeled as cylindricalobjects with hemispherical end-caps in H2O, estimat-ing size obtained from TEM [w, l] � [11, 46] nm. The di-electric function used for Au was obtained from litera-ture values.28 The calculated longitudinal plasmonmode of NRs showed a maximum at a wavelength of824 nm in water. The experimentally observed longitu-dinal plasmon resonance of NRs (804 nm) differs fromthat calculated theoretically because the energy of thelongitudinal mode depends strongly on the end-capgeometry.29 Nonetheless, the calculated scatteringspectra shown in Figure 4 indicate that NSs scatter moreefficiently than NRs for the sizes of nanoparticlesstudied.

Experimentally we observed �34 times strongerscattering from NSs compared to NRs at 785 nm bymeasuring angular scattering distribution (SupportingInformation Figure S2b). The larger scattering efficiencyof NSs relative to NRs studied here correlates with thelarger fluorescence enhancement observed for IR800 byNSs.

In addition to the scattering characteristics of thenanoparticles, increasing the radiative decay rate ofthe fluorophore contributes substantially to fluores-cence enhancement. In a homogeneous solution, fluo-rophores emit light into free space and are observed inthe far field. The observed emission of the fluorophorein the absence of any quenching interactions is de-scribed in terms of quantum yield (Q0) and lifetime (�0).The quantum yield, given by

is the fraction of the excited fluorophores which re-

laxes by radiative decay (�) relative to the total relax-

ation rate (� knr). The observed lifetime is simply the

inverse of the total decay rate of the excited state:

When a fluorophore is in the presence of a metal nano-

particle, the enhanced near field of the nanoparticle in-

creases the amount of light absorbed by the molecule.

In addition, electromagnetic coupling occurs between

the fluorophore and the nanoparticle plasmon, causing

an increase in the radiative decay rate of the molecule

at the emission wavelength by a factor r. As a result,

the effective radiative decay rate is equivalent to r�.

The modified quantum yield (QM) and lifetime (�M) are

then given by

We assume that the nonradiative decay rate, knr, is un-

affected by the presence of the nanoparticles.

The change in radiative lifetime for IR800 conju-

gated to NSs and NRs via HSA was determined ex-

perimentally by monitoring the frequency domain

fluorescence decay in aqueous solution. The lifetime

spectra of IR800, HSA�IR800, NRs�HSA�IR800, and

NSs�HSA�IR800 are shown in Figure 5. The intrinsic

lifetime of the fluorophore was measured to be

�564 ps, which corresponds well to the previously

reported value.25 The lifetime decreased to 427 ps

when IR800 molecules were bound to HSA (Figure

5b). This is due to a small enhancement in quantum

yield of the fluorophore when conjugated to the pro-

tein, attributable to an increased steric stabilization

of IR800 bound to HSA. This reduces the mobility of

the fluorophore in aqueous solution, resulting in

higher stability and a decreased lifetime.30 The life-

time of IR800 bound to the nanoparticles via HSA

was reduced significantly to 121 ps for NRs�HSA�

IR800 and 68 ps for NSs�HSA�IR800. These mea-

sured lifetimes confirm the theoretical prediction of

enhanced quantum yield and diminished lifetime of

fluorophores near metallic surfaces.7

In the frequency domain technique, lifetime mea-

surements are obtained by exciting the fluorescent

sample with intensity-modulated light at a high fre-

quency comparable to ��1 of the sample. Subsequent

to excitation, the fluorescence emission of the sample

follows the same modulation frequency as the excita-

tion, but it is delayed in time relative to the excitation.

Figure 4. Calculated scattering spectra of (i) NSs [r1, r2] �[63, 78] nm, �max-scattering observed at 800 nm and (ii) NRs [w,l] � [11, 46] nm, �max-scattering observed at 824 nm.

Q0 ) ΓΓ + knr

(1)

τ0 ) 1Γ + knr

(2)

QM )γrΓ

γrΓ + knr(3)

τM ) 1γrΓ + knr

(4)

ARTIC

LE

www.acsnano.org VOL. 3 ▪ NO. 3 ▪ 744–752 ▪ 2009 747

The time delay is measured as a phase shift (�) from

which the lifetime (��) is determined:

where � is angular frequency. The fluorescence life-

time data can be evaluated in terms of a single expo-

nential (SE) model or a multiexponential (ME) model. A

SE model is appropriate for samples consisting of a

single fluorophore in a homogeneous environment,

while a ME model describes the fractional contribution

of decay time for each component present in a sample

mixture.31 The ME model for analyzing the intensity de-

cay is given by

where �i are observed lifetimes with amplitude or mo-

lecular fraction �i, such that

ΣiRi ) 1

The amplitude-weighted lifetime is then

The results of fitting to a ME decay analysis are rep-

resented in Table 1. The decay rate for IR800 in aque-

ous media, in the absence of any other component in

the solution, fits a SE decay model. However, the

HSA�IR800 complex required a ME decay model, dem-

onstrating that two different environments are avail-

able to the fluorophore with distinct lifetimes ineach individual environment. About 90% of theIR800 molecules were covalently attached to HSAwith afluorescence decay of 418 ps. The lifetime of the re-maining fraction was similar to that observed forfree IR800 in aqueous media, suggesting that asmall fraction of unbound fluorophores remain inthe HSA�IR800 solution. The ME analysis of thefluorescence decay of IR800 conjugated to nano-particles indicates three distinct environments,which we interpret as IR800 molecules conjugatedto the nanoparticle�HSA, IR800 molecules boundto HSA only, and the free fluorophore in solution.Most of the fluorophores in these samples exhib-ited a significantly reduced lifetime, indicating thatnearly all of the dye molecules were attached to thenanoparticles via the HSA layer. This analysis re-veals that both nanoparticle samples had less than2% of unbound HSA�IR800 complex and less than1% of free fluorophore in the solution mixture.This is attributable to the high affinity of the pro-tein for Au nanoparticle surface as well as the useof a fluorophore with a chemical linker which can

covalently attach to the protein. Therefore,

the nanoparticle�HSA�IR800 system provides a use-

ful complex for determining individual lifetime compo-

nents. The 2R values shown in the last column of Table

1 are the goodness of fit parameter, obtained by fit-

ting calculated values to experimentally obtained pa-

rameters by a nonlinear least-squares deconvulation

method. The 2R values represented here are within

5�10% of the random deviations in the data.31

In order to determine the radiative rate enhance-

ment from the experimentally measured changes in

lifetime, the nonradiative decay rate must be known.

By combining and rearranging eqs 1 and 2, the radia-

tive and nonradiative decay rate can be experimentally

obtained from the lifetime and quantum yield:

Figure 5. Frequency domain fluorescence decay of (a) IR800 (� � 564 ps), (b)HSA�IR800 (� � 427 ps), (c) NRs�HSA�IR800 (� � 121 ps), and (d)NSs�HSA�IR800 (� � 68 ps). The curves referring to phase and modulation areshown with arrows.

τΦ ) 1ω

tan(Φ) (5)

I(t) ) ∑i

Rie-t

τi (6)

⟨τ⟩ ) ΣiRiτi (7)

TABLE 1. Multiexponential Analysis of Intensity Decay ofIR800 with Nanoparticles Showing Molecular Fraction (�),Observed Lifetime (�i, ns), Amplitude-Weighted Lifetime(���, ns), and Goodness of Fit Parameter (�2

R)

sample �i �i ��� �2R

IR8001 0.564

0.564 1.57� �� �

HSA-IR800 0.076 0.5470.427 1.540.924 0.418

� �NRs-HSA-IR800 0.008 0.550 0.121 1.42

0.019 0.4190.973 0.112

NSs-HSA-IR800 0.005 0.568 0.068 1.390.01 0.4210.985 0.065

ART

ICLE

VOL. 3 ▪ NO. 3 ▪ BARDHAN ET AL. www.acsnano.org748

The quantum yield of IR800 (Q) was determined experi-

mentally using indocyanine green (ICG) as a reference

sample with a known quantum yield (QR) of �1% in

aqueous media.32 The NIR excitation�emission profile

of ICG (excitation � 780 nm, emission � 820 nm) is

similar to that of IR800 and is therefore ideal as a refer-

ence sample for quantum yield determination. The

quantum yield was computed by measuring the opti-

cal density of solutions with equivalent concentrations

of IR800 (OD) and ICG (ODR) and by calculating the inte-

grated fluorescence intensity of IR800 (I) and ICG (IR):

where � is the refractive index of sample medium and

�R is the refractive index of the reference medium,

which are equivalent in this experiment (� � �R �

�H2O � 1.33). By using eq 10, the quantum yield of IR800

was determined to be 7% in H2O, and that of the

HSA�IR800 complex was determined to be 11%. Us-

ing the quantum yield and lifetime measured indepen-

dently, eqs 8 and 9 allow determination of the radiative

and nonradiative decay rates of the fluorophore as

shown in Table 2. The quantum yield values obtained

for IR800 in aqueous media as well as HSA�IR800 are

comparable to those reported in the literature.33

The radiative rate enhancement of IR800 induced

by NSs and NRs and the improved quantum yields can

now be calculated by rearranging eqs 3 and 4:

where �M is obtained from Table 1, and knr is the nonra-

diative decay rate for HSA�IR800. We assume that knr

is unaltered from HSA�IR800 for both NSs�HSA�

IR800 and NRs�HSA�IR800 complexes because no

fluorescence quenching was observable and the chemi-

cal environment of the fluorophore was equivalent in

all cases. The nonradiative decay rate does increase

when the chemical environment of the fluorophore is

changed34 from solution phase to being bound toHSA. Table 2 shows that the knr increases when IR800is bound to HSA compared to the free fluorophore insolution. On a metallic surface, the nonradiative decayrate also increases for short fluorophore�metal dis-tances, �4 nm,35,36 since the nonradiative energy trans-fer rate depends on the inverse cube of themolecule�surface separation.35 However, in our nano-structure complexes, HSA provides a spacer layer of �8nm between the fluorophore and the metal nanoparti-cle surface. For this significantly larger metal�moleculedistance, a significant increase in knr due to the metalnanoparticle is not anticipated.6,7 Therefore we assumethat knr is essentially the same for HSA�IR800 in solu-tion and for HSA�IR800 adsorbed sequentially onto ananoshell or nanorod surface. The radiative decay ratesof NRs�HSA�IR800 and NSs�HSA�IR800 and thequantum efficiencies are also shown in Table 2. Thehigh quantum yield of 86% for NSs�HSA�IR800 dem-onstrates that plasmonic enhancement can be used tocreate NIR-fluorescent species with similar intensities asvisible dyes. Although the NRs achieve a lower quan-tum yield of 74% due to the lower scattering efficiencyof the particles, this is still a very high value for a NIR flu-orophore, and these structures would certainly also beuseful as markers in fluorescence-based bioimagingwhere the smaller physical size of the NRs-based nano-particle complex would be desired.

The precise relative contribution of each process re-sponsible for IR800 fluorescence enhancement, includ-ing absorption enhancement, scattering enhancement,and radiative decay rate enhancement, is difficult to de-termine since these processes are interdependent. Nev-ertheless, for the experimental parameters and nano-particle geometries discussed here, the scatteringefficiency of a nanoparticle appears to provide themost important mechanism for improving the quan-tum yield of a fluorophore. NRs predominantly enhancethe emission of the fluorophore by absorption en-hancement, owing to the high-intensity near field re-sulting from the longitudinal plasmon resonance. How-ever, due to the significant difference in scatteringcross sections of NSs and NRs, it is apparent that NSs in-crease the coupling efficiency of the fluorescence emis-sion to the far field more efficiently than NRs. This ex-plains the 40-fold fluorescence enhancement observedfor IR800 bound to NSs compared to the 9-fold en-hancement for IR800 bound to NRs. The radiative de-cay rate enhancement of the fluorophore is dependenton both the scattering efficiency as well as the absorp-tion efficiency of nanoparticles. This explains why NRsenhance the quantum yield of IR800 by 74% as well asdecrease the fluorophore’s lifetime considerably.

CONCLUSIONSIn conclusion, we have examined the fluorescence

enhancement of IR800 conjugated to NSs and NRs,

Γ )Q0

τ0(8)

knr )1τ0

- Γ (9)

Q ) QRI

IR

ODODR

ηηR

(10)

TABLE 2. Quantum Yield (QY), Radiative Decay Rate (�),and Nonradiative Decay Rate (knr) of IR800, HSA�IR800,and HSA�IR800 with Nanoparticles

sample QY � knr

IR800 0.07 1.241 � 108 1.648 � 109

HSA�IR800 0.11 2.576 � 108 2.084 � 109

NRs�HSA�IR800 0.74 6.180 � 109 2.084 � 109

NSs�HSA�IR800 0.86 1.262� 1010 2.084 � 109

γrΓ ) 1τM

- knr (11)

QM ) γrΓ·τM (12)

ARTIC

LE

www.acsnano.org VOL. 3 ▪ NO. 3 ▪ 744–752 ▪ 2009 749

and we have shown that both NSs and NRs lead tolarge increases in quantum yield relative to the iso-lated fluorophore. We have observed that NSs are moreefficient in improving the emissive properties of a fluo-rophore due to their significant scattering cross sectionat the emission wavelength of the fluorophore. Addi-tionally, the near-field response of NSs gives rise to aconsiderable enhancement in the absorption and theradiative decay rate of IR800, resulting in 40-fold en-hancement and 86% quantum yield. IR800 molecules

bound to NRs demonstrate a 9-fold emission enhance-ment and a 74% quantum yield, attributable to the highlocal field enhancement at the longitudinal plasmonwavelength. Utilizing Au nanoparticles with appropri-ate geometry and dimensions for emission enhance-ment is a useful strategy for enhancing the detectionsensitivity of low-quantum-yield fluorescent emitters.This approach is also potentially valuable in biomedi-cal imaging and, moreover, can be conveniently gener-alized to enhance other fluorescent media.

METHODSSection I. Nanoparticle Fabrication. All chemicals were purchased

from Sigma Aldrich, and IR800 NHS ester was purchased from Li-cor Biosceinces. Au nanoshells (NSs) [r1, r2] � [63, 78] nm werefabricated by seed-mediated electroless plating of Au onto silicaspheres as previously reported.37 Briefly, monodisperse silicananospheres of 63 � 2 nm radii were synthesized by the hydrol-ysis of tetraethylorthosilicate (TEOS) in basic solution via the Sto-ber method38 and functionalized with (3-aminopropyl)triethoxysilane (APTES) overnight. These amine-terminated silicananospheres were decorated with small gold colloid (2�3 nm)prepared by the method reported by Duff et al.39 A continuousgold shell is grown around the silica nanospheres by reducinggold from a 1% solution of HAuCl4 onto the attached small col-loid in the presence of formaldehyde. The fabricated NSs werecentrifuged several times and finally redispersed in sodium phos-phate buffer (Na3PO4, pH � 8.0).

Au nanorods (NRs) [w, l] � [11, 46] nm were fabricated bythe method reported by Murphy and co-workers.40 Briefly, theseed solution was prepared by gently mixing 7.5 mL of 0.1 Mcetyltrimethylammonium bromide (CTAB) with 0.25 mL of 0.01M HAuCl4; 0.6 mL of ice-cold 0.01 M NaBH4 was injected into thesolution and mixed rapidly. The solution turns from a golden yel-low color to a pale brown color. It is essential that the tempera-ture of NaBH4 is maintained at �0 °C prior to mixing with CTABand HAuCl4 for proper growth of nanorods. The seed solutionwas then stored at 27 °C until further use. The growth solutionwas prepared by gently mixing 47.5 mL of 0.1 M CTAB, 2 mL of0.01 M HAuCl4, 0.3 mL of 0.01 M AgNO3, and 0.32 mL of 0.1 Mascorbic acid in the order mentioned. Addition of ascorbic acidturns the solution from golden yellow to colorless; 0.25 mL of theseed solution was added to the growth solution and mixed gen-tly, and then the solution was stored at 27 °C for 3 h without agi-tation. Within 20 min, the solution changes from colorless to pinkand eventually changes to dark pinkish brown after 3 h. TheNRs were washed twice at 8000 rpm to remove excess CTABand finally redispersed in 10 mL of Na3PO4 buffer.

Section II. Protein and Fluorophore Conjugation. HSA was conju-gated to IR800 by mixing equal volumes of 5 �M HSA aqueoussolutions in Na3PO4 buffer and 30 �M IR800 aqueous solution atroom temperature for 5 h, in the dark. The protein�flurophorecomplex was then dialyzed in Na3PO4 buffer for 24 h at roomtemperature protected from light, in a 3500 MW dialysis bag(MW of HSA is 67 kDA and MW of IR800 is 1166 g/mol) to re-move excess dye. Absorbance peaks of the protein�dye com-plex (�max � 782 nm) and protein (�max � 280 nm) were moni-tored to account for free-dye removal. After sufficient removal ofunbound dye, the ratio of fluorophore to protein concentration(NIR800/NHSA) of the complex is determined by utilizing the ab-sorbance of fluorophore at 782 nm (AIR800), molar extinction co-efficient of IR800 (�IR800), absorbance of HSA at 280 nm (AHSA), mo-lar extinction coefficient of HSA (�HSA) given by

NIR800/NHSA ) [AIR800/εIR800]/[AHSA/εHSA]

where �IR800 � 240 000 M�1 cm�1 and �HSA � 42 864 M�1 cm�1.The final protein concentration was calculated using the

molecular weight (MWHSA) and dilution factor (df) of originalsolution given by

NHSA ) [AHSA/εHSA] × MWHSA × df

The final protein concentration was �4 �M, and fluorophoreconcentration was �10 �M.

Section III. Protein�Fluorophore Complex Binding to Nanoparticles.The protein�fluorophore complex obtained was conjugatedto the nanoparticles by adding 1 mL of HSA�IR800 solutionin Na3PO4 buffer to 10 mL nanoshells and 10 mL nanorodsdispersed in Na3PO4 buffer, respectively, under constant stir-ring. After 7 h conjugation at room temperature, in the dark,the nanoparticles were centrifuged in small aliquots of 500�L to remove excess dye and resuspended in Na3PO4 buffer.The absorbance of the supernatant was closely monitored toaccount for unbound fluorophore concentration in the super-natant (9.55�9.65 �M). The concentration of the superna-tant was then subtracted from the initial concentration ofHSA�IR800, which was added to the nanoparticles to calcu-late the amount of HSA�IR800 on the nanoparticles. The fi-nal NSs concentration after resuspension in buffer was at�108 particles/mL, and final NRs concentration was at �109

particles/mL. The surface area available for the HSA�IR800complex to bind to the nanoparticles was normalized (seebelow), and hence the concentration of fluorophore wasequivalent for both the NSs�HSA�IR800 andNRs�HSA�IR800, �400 � 50 nM.

Instrumentation and Modeling. The nanoparticles were character-ized by obtaining transmission electron microscope (TEM) im-ages using a JEOL JEM-2010 TEM and absorbance measurementsusing a Varian Cary 5000 UV�vis�NIR spectrometer. Fluores-cence emission spectra and frequency domain lifetime measure-ments were obtained using Jobin Yvon Fluorolog 3, and thesamples were excited at 780 nm. The dynamic range of the Flu-orolog 3 for lifetime measurements is 0�300 MHz. Angular scat-tering distribution was obtained using an in-house angle-resolved fluorescence emission apparatus.

The theoretical scattering intensity of NRs was calculated byusing the finite element method (FEM) (as implemented byCOMSOL Multiphysics).

Section IV. Calculation of Surface Area Available for Protein�DyeComplex To Bind to Nanoparticles. The amount of protein and fluoro-phore bound to the nanoparticles was quantified by assumingthat IR800 binds to HSA similarly for both NSs and NRs. The avail-able surface area for fluorophore binding is kept equivalent forboth nanoparticle solutions. The concentration (CNS) and extinc-tion coefficient (�NS) for NSs were calculated utilizing Mietheory:

σNS ) 9.6072 × 10-10 cm2 for NSs [r1, r2] ) [63, 78] nm

CNS ) ln(10) × absorbance/σ × cuvette path length

) 2.303 × 0.052/9.6072 × 10-10 cm2 × 1 cm

) 1.25 × 108 particles/mL

ART

ICLE

VOL. 3 ▪ NO. 3 ▪ BARDHAN ET AL. www.acsnano.org750

the surface area of one NS (SANS) ) 4π[r22 - r1

2] )

2.855 × 10-14 m2

the surface area of NSs solution ) (SANS) × CNS )

3.568 × 10-6 m2/mL

The extinction coefficient (�NR) for NRs [w, l] � [11, 46] nmresonant at �800 nm has been reported previously as �NR � 4� 10�11 cm2.41 The concentration of NRs (CNR) were calculatedutilizing Gan’s theory:

CNR ) 2 × absorbance/σ × cuvette path length

) 2 × 0.043/4 × 10-11 cm2 × 1 cm

) 2.154 × 109 particles/mL

the surface area of one NR (SANR) ) 2πr2 + 2πrl )

1.727 × 10-15 m2

the surface area of NRs solution ) (SANR) × CNR )

3.719 × 10-6 m2/mL

However, experimentally, the available nanoparticle surface areafor fluorophore binding would vary slightly due to nonuniformsize distribution of nanoparticles in solution.

Section V. Calculation of Number of Protein�Fluorophore MoleculesBound to the Nanoparticles. The diameter of a HSA molecule is ap-proximately �8 � 3 nm, and we assume the diameter of a IR800molecule is roughly �2 nm, resulting in a hydrodynamic diam-eter of �12 nm for the protein�fluorophore complex. We havedetermined the number of fluorophore molecules per proteinmolecule by taking a ratio of their surface area:

the surface area of one HSA molecule (SAHSA) ) 4πr2 )

4π(4 × 10-9)2 ) 2.011 × 10-16 m2

the surface area of one IR800 molecule (SAIR800) )

4π(1 × 10-9)2 ) 1.256 × 10-17 m2

the no. of IR800 molecules per HSA molecule ) SAHSA/SAIR800 ≈16 dye molecules

We have also calculated the number of molecules ofprotein�fluorophore complex per nanoparticle, for both NSs aswell as NRs:

the surface area of one HSA-IR800 molecule (SAHSA-IR800) )

4π(6 × 10-9)2

) 4.523 × 10-16 m2

the no. of HSA-IR800 molecules per NS ) SANS/SAHSA-IR800

) 2.855 × 10-14 m2/4.523 × 10-16 m2

≈ 63 HSA-IR800 molecules

the no. of HSA-IR800 molecules per NR ) SANR/SAHSA-IR800

) 1.727 × 10-15 m2/4.523 × 10-16 m2

≈ 4 HSA-IR800 molecules

However, as described above, the surface area available for theHSA�IR800 complex to bind to NSs and NRs has beennormalized by adjusting the nanoparticle concentration. Thisprovides almost equivalent amounts of protein�dye moleculesto bind to each nanoparticle suspension.

Acknowledgment. We gratefully acknowledge the Robert A.Welch Foundation (C-1220), Air Force Office of Scientific Re-search Grant (F49620-03-C-0068), and Multidisciplinary Univer-sity Research Initiative (W911NF-04-01-0203) for financial sup-

port. We thank Prof. Bruce Johnson, Lisa Brown, Britt Lassiter,Nikolay Mirin, and Surbhi Lal for helpful discussions, RebekahDrezek for fluorescence emission measurements, and Eva M.Sevick-Muraca for lifetime measurements.

Supporting Information Available: Spectra showing plasmonresonance shift when nanoparticles are coated with HSA�IR800;schematic diagram of home-built angle-resolved fluorescenceemission apparatus; angle-resolved scattering spectra of NSs andNRs; fluorescence spectra of control: nanostructures without flu-orophore; fluorescence spectra of nanostructure�fluorophoreconjugates excited at 720 nm. This material is available free ofcharge via the Internet at http://pubs.acs.org.

REFERENCES AND NOTES1. Hillman, E. M. C.; Moore, A. All-Optical Anatomical Co-

registration for Molecular Imaging of Small Animals UsingDynamic Contrast. Nat. Photonics 2007, 1, 526–530.

2. Kovar, J. L.; Simpson, M. A.; Schutz-Geschwender, A.; Olive,D. M. A Systematic Approach to the Development ofFluorescent Contrast Agents for Optical Imaging of MouseCancer Models. Anal. Biochem. 2007, 367, 1–12.

3. Weissleder, R. A Clearer Vision For in vivo Imaging. Nat.Biotechnol. 2001, 19, 316–317.

4. Anger, P.; Bharadwaj, P.; Novotny, L. Enhancement andQuenching of Single-Molecule Fluorescence. Phys. Rev.Lett. 2006, 96, 1130021–1130024.

5. Drexhage, K. H. Influence of a Dielectric Interface OnFluorescence Decay Time. J. Lumin. 1970, 1, 693–701.

6. Tam, F.; Goodrich, G. P.; Johnson, B. R.; Halas, N. J.Plasmonic Enhancement of Molecular Fluorescence. NanoLett. 2007, 7, 496–501.

7. Bardhan, R.; Grady, N. K.; Halas, N. J. Nanoscale Control ofNear-Infrared Fluorescence Enhancement Using AuNanoshells. Small 2008, 4, 1716–1722.

8. Lakowicz, J. R. Radiative Decay Engineering 5: Metal-Enhanced Fluorescence and Plasmon Emission. Anal.Biochem. 2005, 337, 171–194.

9. Anker, J. N.; Hall, W. P.; Lyandres, O.; Shah, N. C.; Zhao, J.;Van Duyne, R. P. Biosensing With Plasmonic Nanosensors.Nat. Mater. 2008, 7, 442–453.

10. Liao, H.; Nehl, C. L.; Hafner, J. H. Biomedical Applications ofPlasmon Resonant Metal Nanoparticles. Nanomedicine2006, 1, 201–208.

11. Chance, R. R.; Prock, A.; Silbey, R. Lifetime of an EmittingMolecule Near a Partially Reflecting Surface. J. Chem. Phys.1974, 60, 2744–2748.

12. Wang, H.; Brandl, D. W.; Nordlander, P.; Halas, N. J.Plasmonic Nanostructures: Artificial Molecules. Acc. Chem.Res. 2007, 40, 53–62.

13. Lee, K. S.; El-Sayed, M. A. Gold and Silver Nanoparticles inSensing and Imaging: Sensitivity of Plasmon Response toSize, Shape, and Metal Composition. J. Phys. Chem. B 2006,110, 19220–19225.

14. Hao, F.; Nehl, C. L.; Hafner, J. H.; Nordlander, P. PlasmonResonances of a Gold Nanostar. Nano Lett. 2007, 7,729–732.

15. Wang, H.; Goodrich, G. P.; Tam, F.; Oubre, C.; Nordlander,P.; Halas, N. J. Controlled Texturing Modifies the SurfaceTopography and Plasmonic Properties of Au Nanoshells. J.Phys. Chem. B 2005, 109, 11083–11087.

16. Tam, F.; Chen, A. L.; Kundu, J.; Wang, H.; Halas, N. J.Mesoscopic Nanoshells: Geometry-Dependent PlasmonResonances Beyond the Quasistatic Limit. J. Chem. Phys.2007, 127, 204703.

17. Nikoobakht, B.; El-Sayed, M. A. Preparation and GrowthMechanism of Gold Nanorods (NRs) Using Seed-MediatedGrowth Method. Chem. Mater. 2003, 15, 1957–1962.

18. Prodan, E.; Radloff, C.; Halas, N. J.; Nordlander, P.Hybridization Model for the Plasmon Response ofComplex Nanostructures. Science 2003, 302, 419–422.

19. Sokolov, K.; Chumanov, G.; Cotton, T. M. Enhancement ofMolecular Fluorescence near the Surface of Colloidal MetalFilms. Anal. Chem. 1998, 70, 3898–3905.

ARTIC

LE

www.acsnano.org VOL. 3 ▪ NO. 3 ▪ 744–752 ▪ 2009 751

20. He, X. M.; Carter, D. C. Atomic Structure and Chemistry ofHuman Serum Albumin. Nature 1992, 358, 209–215.

21. Tkachenko, A. G.; Xie, H.; Liu, Y.; Coleman, D.; Ryan, J.;Glomm, W. R.; Franzen, S.; Feldheim, D. L. CellularTrajectories of Peptide-Modified Gold Particle Complexes:Comparison of Nuclear Localization Signals and PeptideTransduction Domains. Bioconjugate Chem. 2004, 15, 482–490.

22. Rosi, N. L.; Mirkin, C. A. Nanostructures in Biodiagnostics.Chem. Rev. 2005, 105, 1547–1562.

23. Brewer, S. H.; Glomm, W. R.; Johnson, M. C.; Knag, M. K.;Franzen, S. Probing BSA Binding to Citrate-Coated GoldNanoparticles and Surfaces. Langmuir 2005, 21,9303–9307.

24. Patel, N.; Davies, M. C.; Heaton, R. J.; Roberts, C. J.; Tendler,S. J. B.; Williams, P. M. A Scanning Probe Microscopy Studyof the Physisorption and Chemisorption of ProteinMolecules onto Carboxylate Terminated Self-AssembledMonolayers. Appl. Phys. A: Mater. Sci. Process. 1998, 66,S569–S574.

25. Houston, J. P.; Ke, S.; Wang, W.; Li, C.; Sevick-Muraca, E. M.Quality Analysis of in vivo Near-Infrared Fluorescence andConventional Gamma Images Acquired Using a Dual-Labeled Tumor-Targeting Probe. J. Biomed. Opt. 2005, 10,0540101–05401011.

26. Prodan, E.; Lee, A.; Nordlander, P. The Effect of a DielectricCore and Embedding Medium on the Polarizability ofMetallic Nanoshells. Chem. Phys. Lett. 2002, 360, 325–332.

27. Link, S.; Mohamed, M. B.; El-Sayed, M. A. Simulation of theOptical Absorption Spectra of Gold Nanorods as aFunction of Their Aspect Ratio and the Effect of theMedium Dielectric Constant. J. Phys. Chem. B 1999, 103,3073–3077.

28. Johnson, P. B.; Christy, R. W. Optical Constants of theNoble Metals. Phys. Rev. B 1972, 6, 4370–4379.

29. Prescott, S. W.; Mulvaney, P. Gold Nanorod ExtinctionSpectra. J. Appl. Phys. 2006, 99, 1235041–1235047.

30. Valdes-Aguilera, O.; Cincotta, L.; Foley, J.; Kochevar, I. E.Photobleaching of a Cyanine Dye in Solution and inMembranes. Photochem. Photobiol. 1987, 45, 337–344.

31. Lakowicz, J. R. Time-Domain Lifetime Measurements. InPrinciples of Fluorescence Spectroscopy, 2nd ed.; KluwerAcademic/Plenum: New York, 1999; pp 95�184.

32. Soper, S. A.; Mattingly, Q. L. Steady-State and PicosecondLaser Fluorescence Studies of Nonradiative Pathways inTricarbocyanine Dyes: Implications to the Design of Near-IR Fluorochromes with High Fluorescence Efficiencies.J. Am. Chem. Soc. 1994, 116, 3144–3152.

33. Ohnishi, S.; Lomnes, S. J.; Laurence, R. G.; Gogbashian, A.;Mariani, G.; Frangioni, J. V. Organic Alternatives toQuantum Dots for Intraoperative Near-InfraredFluorescent Sentinel Lymph Node Mapping. Mol. Imaging2005, 4, 172–181.

34. Lakowicz, J. R. Radiative Decay Engineering: Biophysicaland Biomedical Applications. Anal. Biochem. 2001, 298, 1–24.

35. Waldeck, D. H.; Alivisatos, A. P.; Harris, C. B. NonradiativeDamping of Molecular Electronic Excited States by MetalSurfaces. Surf. Sci. 1985, 158, 103–125.

36. Soller, T.; Ringler, M.; Wunderlich, M.; Klar, T. A.; Feldmann,J.; Josel, H.-P.; Markert, Y.; Nichtl, A.; Kurzinger, K. Radiativeand Nonradiative Rates of Phosphors Attached to GoldNanoparticles. Nano Lett. 2007, 7, 1941–1946.

37. Oldenburg, S. J.; Averitt, R. D.; Westcott, S. L.; Halas, N. J.Nanoengineering of Optical Resonances. Chem. Phys. Lett.1998, 288, 243–247.

38. Stober, W.; Fink, A.; Bohn, E. J. Controlled Growth ofMonodisperse Silica Spheres in the Micron Size Range. J.Colloid Interface Sci. 1968, 26, 62–69.

39. Duff, D. G.; Baiker, A.; Edwards, P. P. A New Hydrosol ofGold Clusters. 1. Formation and Particle Size Variation.Langmuir 1993, 9, 2301–2309.

40. Sau, T. K.; Murphy, C. J. Seeded High Yield Synthesis ofShort Au Nanorods in Aqueous Solution. Langmuir 2004,20, 6414–6420.

41. Gulati, A.; Liao, H.; Hafner, J. H. Monitoring Gold NanorodSynthesis by Localized Surface Plasmon Resonance. J.Phys. Chem. B 2006, 110, 22323–22327.

ART

ICLE

VOL. 3 ▪ NO. 3 ▪ BARDHAN ET AL. www.acsnano.org752